Founders Dream, page-136

BS is an unsupervised machine learning software, it implements JAST as its learning algorithms; supervised learning needs big dataset for training, BS doesnt need a big dataset to train the machine learning.

HA improves BS throughput by 6 times;

akida is neuralmorphic system on chip, implements JAST unsupervised learning rules for pattern recognition. Akida needs CPU to run the deep learning(as the process described in LadyL post, brn implements third generation SNN deep learning, this improves speed and saves power) rules.

That's my understanding, may not correct.